Mining equipment online after-sales service method and system based on knowledge graph

Abstract

The application provides a mine-used equipment online after-sales service method and system based on a knowledge graph. The method comprises the following steps: acquiring multiple types of knowledge related to mine-used equipment after-sales service, and processing the multiple types of knowledge to generate a knowledge graph; constructing an after-sales expert online operation and maintenance platform based on the knowledge graph, answering after-sales problems raised by users online, and updating the knowledge graph according to information generated in the problem answering process; constructing a user interest model according to the real-time updated knowledge graph, and gradually generating an operation and maintenance knowledge recommendation list for the user according to the user interest model; and in the case that the user raises an intelligent question and answer request, carrying out multi-round dialogue with the user based on the operation and maintenance knowledge recommendation list and aiming at the current after-sales problem. The method provides intelligent online operation and maintenance functions based on artificial intelligence technologies such as the knowledge graph, improves the accuracy and timeliness of mine-used equipment after-sales service, and reduces service costs.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of coal mine equipment operation and maintenance, and particularly relates to a mine-used equipment online after-sales service method and system based on a knowledge graph. BACKGROUND

[0002] With the development of intelligent mining of coal mines, the types and quantities of mine-used equipment arranged in underground coal mines gradually increase. In the actual production process of mines, problems such as hardware failure and software failure of installed equipment may occur, which need to be handled by after-sales maintenance.

[0003] In related technologies, when after-sales service is performed on a faulty equipment, a relevant mine-used equipment after-sales service personnel is usually contacted to go to the scene to troubleshoot problems. However, the mine-used equipment that needs to be operated and maintained may be scattered in various locations in a large area, and the number of mine-used equipment to be operated and maintained is large, which needs to be checked one by one by the after-sales service personnel. Therefore, the manual after-sales service mode in related technologies has problems such as low after-sales service efficiency, high service cost, and unprofessional service. SUMMARY

[0004] The present application aims to at least solve one of the technical problems in related technologies to some extent.

[0005] To this end, a first purpose of the present application is to propose a mine-used equipment online after-sales service method based on a knowledge graph, which collects mine-end problems through the construction of an online operation and maintenance service system, realizes functions such as online expert answers, online operation and maintenance intelligent search, and knowledge base intelligent recommendation, and performs closed-loop management on after-sales service, can feedback the after-sales service progress to the customer in time, and continuously enrich the operation and maintenance knowledge base, summarize operation and maintenance problems, and provide data support for further improving the service quality.

[0006] A second purpose of the present application is to propose a mine-used equipment online after-sales service system based on a knowledge graph.

[0007] A third purpose of the present application is to propose a non-transitory computer readable storage medium.

[0008] To achieve the above purpose, a first aspect embodiment of the present application proposes a mine-used equipment online after-sales service method based on a knowledge graph, which comprises the following steps:

[0009] acquiring multiple types of knowledge related to mine-used equipment after-sales service, wherein the multiple types of knowledge include multiple elements of mine-used equipment online operation and maintenance service, and processing the multiple types of knowledge to generate a knowledge graph;

[0010] An online operation and maintenance platform for after-sales experts is constructed based on the knowledge graph, the online operation and maintenance platform is used to answer the after-sales problems raised by the user online, and the knowledge graph is updated according to the information generated in the process of answering the problems;

[0011] A user interest model is constructed according to the real-time updated knowledge graph, a user demand similarity matrix is calculated according to the user interest model, a user after-sales service demand score is predicted according to the user demand similarity matrix, and an operation and maintenance knowledge recommendation list for the user is generated according to the after-sales service demand score;

[0012] In the case that the user raises an intelligent question and answer request, a multi-round dialogue for the current question and answer request is carried out between the user and the operation and maintenance knowledge recommendation list.

[0013] Optionally, in an embodiment of the present application, the processing of the multi-type knowledge to generate a knowledge graph includes: knowledge extraction is performed on the multi-type knowledge to obtain entity information, relationship information and attribute information, and a preliminary knowledge representation is generated based on the extracted information; entity alignment is performed on the preliminary knowledge representation to obtain a standard knowledge representation, and quality evaluation is performed on the standard knowledge representation, and the standard knowledge representation that passes the quality evaluation is taken as the knowledge graph.

[0014] Optionally, in an embodiment of the present application, the multiple elements of the online operation and maintenance service include: after-sales service engineers, mine end users, mine equipment, equipment software and operation and maintenance work, and the online answering of the after-sales problems raised by the user through the online operation and maintenance system includes: receiving the after-sales problems raised by the mine end users, determining a candidate engineer set corresponding to the after-sales problems based on the knowledge graph, and sending the after-sales problems to each candidate engineer; determining a target after-sales service engineer for answering the after-sales problems from the candidate engineer set through a recognition or superior supervisor assignment; and sending the problem solving scheme fed back by the target after-sales service engineer to the mine end user for offline processing.

[0015] Optionally, in an embodiment of the present application, after the problem solving scheme fed back by the target after-sales service engineer is sent to the mine end user for offline processing, it further includes: obtaining the problem processing result returned by the mine end user; in the case that the after-sales problem is not solved, sending the problem processing result of this round to the target after-sales service engineer for re-answering, and counting the number of problem answering, in the case that the number of problem answering exceeds a preset threshold, replacing the target after-sales service engineer; in the case that the after-sales problem is solved, obtaining the score of the target after-sales service engineer given by the mine end user, and adjusting the rating of the target after-sales service engineer according to the score and the tracking record of the superior supervisor.

[0016] Optionally, in an embodiment of the present application, the constructing the user interest model according to the real-time updated knowledge graph comprises: determining an interest field of the user, collecting data related to the interest field from the knowledge graph, the data being represented as entity information, relationship information and attribute information; performing vectorization representation on the entity information, the relationship information and the attribute information in the data through a TransR model, and calculating a user interest vector according to the vector representation of the obtained triplets; creating an initial interest model, and training the initial interest model through the user interest vector; evaluating and optimizing the trained interest model to generate the user interest model.

[0017] Optionally, in an embodiment of the present application, the calculating the user demand similarity matrix according to the user interest model comprises: representing the user interest model as a user vector, the user vector being used to determine the similarity between the user and the after-sales service demand; calculating the user demand similarity matrix by calculating the similarity between the user vectors corresponding to different users; and the predicting the after-sales service demand score of the user according to the user demand similarity matrix comprises: predicting the score of the un-scored after-sales service demand of the target user through a weighted average calculation manner, wherein the weight in the weighted average is the product of the score of the un-scored after-sales service demand of other different users and the corresponding user demand similarity matrix, the corresponding user demand similarity matrix being the user demand similarity matrix between other different users and the target user.

[0018] Optionally, in an embodiment of the present application, the performing entity alignment on the preliminary knowledge representation comprises: preprocessing entity information in different data sources, and extracting an entity feature vector from the preprocessed entity information; calculating the similarity between different entity information according to the entity feature vector, and performing entity matching according to the similarity; and the performing quality evaluation on the standard knowledge representation comprises: performing entity link accuracy evaluation, relationship integrity evaluation, query response efficiency evaluation, data update and maintenance evaluation on the standard knowledge representation.

[0019] To achieve the above purpose, a second aspect of the present application further proposes a mine-used equipment online after-sales service system based on a knowledge graph, comprising the following modules:

[0020] An acquisition module is configured to acquire multiple types of knowledge related to mine-used equipment after-sales service, wherein the multiple types of knowledge comprise multiple elements of mine-used equipment online operation and maintenance service, and the multiple types of knowledge are processed to generate a knowledge graph.

[0021] The first operation and maintenance service module is configured to construct an online operation and maintenance platform of an after-sales expert based on the knowledge graph, answer the after-sales problems raised by the user online through the online operation and maintenance platform, and update the knowledge graph according to the information generated in the process of answering the problems;

[0022] The generating module is configured to construct a user interest model according to the real-time updated knowledge graph, calculate a user demand similarity matrix according to the user interest model, predict a score of the after-sales service demand of the user according to the user demand similarity matrix, and generate an operation and maintenance knowledge recommendation list for the user according to the score of the after-sales service demand;

[0023] The second operation and maintenance service module is configured to, in the case that the user raises an intelligent question and answer request, carry out a multi-round dialogue with the user for the current question and answer request based on the operation and maintenance knowledge recommendation list.

[0024] To achieve the above-mentioned embodiments, the third aspect of the present application further provides a non-transitory computer-readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the method for online after-sales service of mining equipment based on a knowledge graph in the first aspect of the present application.

[0025] The technical scheme provided by the embodiments of the present application at least brings the following beneficial effects: the present application provides intelligent online operation and maintenance function based on knowledge graph and deep learning and other artificial intelligence technologies. On the one hand, the model is trained by using massive data, and based on the powerful aggregation information and reasoning ability of neural network, a recommendation system based on graph neural network and knowledge graph is designed, which can effectively improve the hit rate of recommendation. And based on the knowledge reasoning knowledge base question and answer, professional technical support can be provided to the mine personnel, the software intelligent question and answer is realized, the professional knowledge for solving problems independently is provided, which is conducive to better safety production, timely and rapid solution of daily problems, and provision of standardized after-sales service process. On the other hand, the online operation and maintenance system of the after-sales expert is established, the communication bridge between the customer and the expert is established, the online operation and maintenance service platform of the expert is established, the quality of the after-sales service is improved, and the actual generated operation and maintenance data is summarized and integrated. According to the actual experience data, the model can be further optimized and the operation and maintenance knowledge base can be enriched. Therefore, the present application solves the problems of complex after-sales service process, unprofessional after-sales service, untimely answer and high after-sales cost, provides an online operation and maintenance closed-loop management process specification, provides intelligent online operation and maintenance function based on knowledge graph and other artificial intelligence technologies, improves the accuracy and timeliness of the after-sales service of the mining equipment, and reduces the service cost.

[0026] Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art upon examination of the following and / or can be learned by practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0027] The above and / or additional aspects and advantages of the present application will become apparent and more readily appreciated from the following description of the embodiments, taken in conjunction with the references to the following drawings, of which:

[0028] Figure 1 A flow chart of a method for online after-sales service of mining equipment based on a knowledge graph according to an embodiment of the present application;

[0029] Figure 2 A schematic diagram of a construction principle of a knowledge graph according to an embodiment of the present application;

[0030] Figure 3 A schematic diagram of a structure of an online after-sales expert operation and maintenance platform according to an embodiment of the present application;

[0031] Figure 4 A schematic diagram of an online operation and maintenance closed-loop management process according to an embodiment of the present application;

[0032] Figure 5 A schematic diagram of an online operation and maintenance intelligent recommendation process according to an embodiment of the present application;

[0033] Figure 6 A flow chart of a construction method of a user interest model according to an embodiment of the present application;

[0034] Figure 7 A schematic diagram of a structure of an online after-sales service system of mining equipment based on a knowledge graph according to an embodiment of the present application. DETAILED DESCRIPTION

[0035] Embodiments of the present application are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below by reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0036] It should be noted that in the actual production process of the mine, the troubleshooting process of various faults of the mining equipment is complex, and the processing time is long. Through telephone communication, it is often not clear to speak, and the after-sales service personnel need to check the equipment on site. If the equipment failure, software failure, etc. can be recorded in the form of pictures, videos and descriptions, it will be convenient for the after-sales service personnel to troubleshoot problems, thereby saving communication time and optimizing the after-sales service process. Moreover, if the problems encountered by the mine personnel in the process of safe production can be solved quickly with the support of professional knowledge without the on-site operation of the after-sales service personnel, the after-sales operation amount will be greatly reduced. Because the on-site after-sales service personnel often shoulder the after-sales service work of the mine equipment scattered in various places in a large area, sometimes they have to consume a lot of time and economic costs to troubleshoot a device problem. If there is a professional knowledge base, the reuse rate of operation and maintenance knowledge is increased, professional answers are provided for mine personnel, the business trip frequency of engineering personnel is greatly reduced, and the after-sales service cost is reduced.

[0037] With the advent of the "Internet +" era, a large amount of information will be generated in various industries every day, and will show an explosive growth trend. The knowledge graph technology can sort these information into standardized knowledge, build a knowledge base, and form reusable explicit knowledge. Therefore, the application builds a knowledge base system based on knowledge graph and other artificial intelligence technologies, provides a knowledge accumulation and inheritance method for online after-sales service of mining equipment. By uniformly constructing a multi-element heterogeneous knowledge base system including a question and answer library, a multi-round dialogue library, a scene skill library and a mining equipment after-sales knowledge base, and combining with hundreds of millions of knowledge data such as system knowledge, industry knowledge and enterprise knowledge, semantic understanding, question and answer reasoning and other intelligent services are provided.

[0038] The application will be described in detail below with reference to the accompanying drawings.

[0039] Figure 1 A flowchart of a mining equipment online after-sales service method based on a knowledge graph according to an embodiment of the application is shown in Figure 1 The method comprises the following steps:

[0040] In step S101, a plurality of types of knowledge related to the after-sales service of the mining equipment are acquired, wherein the plurality of types of knowledge include a plurality of elements of the online operation and maintenance service of the mining equipment, and the plurality of types of knowledge are processed to generate a knowledge graph.

[0041] Specifically, the application constructs a mine-used equipment after-sales online service system based on knowledge graph technology, which can realize two functions, i.e. contains two subsystems. The first aspect is an after-sales expert online operation and maintenance platform, and the other aspect is a personalized after-sales knowledge service recommendation system. The data generated by the two subsystems in actual operation can be accumulated as data, forming an online operation and maintenance closed-loop management process.

[0042] Among them, the mine-used equipment after-sales online service system with the above functions needs to use knowledge graph, so how to construct the knowledge graph model directly affects the accuracy of the provided after-sales service. When constructing the knowledge graph, the application first acquires various types of knowledge related to mine-used equipment after-sales service.

[0043] For example, collect the knowledge of multiple elements involved in online operation and maintenance services required for constructing the after-sales expert online operation and maintenance platform, including after-sales service engineers, mine end users, mine-used equipment, equipment software and operation and maintenance work. The application first collects several categories of data related to mine-used equipment, software, services and customers, and can also collect data including the mine area where the equipment and users are located, equipment historical operation knowledge and equipment troubleshooting operation knowledge, i.e. the application can comprehensively collect various types of knowledge related to mine-used equipment after-sales service to improve the comprehensiveness and accuracy of the subsequent provided after-sales service.

[0044] Further, the acquired multiple types of knowledge are processed to generate a knowledge graph for mine-used equipment online after-sales service.

[0045] In an embodiment of the application, as shown in Figure 2 When generating the knowledge graph, first, the multiple types of raw data collected are integrated and knowledge is extracted, the elements of the expert online operation and maintenance service platform including engineers, customers, equipment, software and operation and maintenance work are integrated together, and the multiple types of knowledge after integration are preliminarily preprocessed to remove abnormal data. Then, the semi-structured and unstructured data in the collected multiple types of knowledge are extracted, the entity information, relationship information and attribute information are extracted, the preliminary knowledge representation is generated based on the extracted information. The preliminary knowledge representation is aligned to obtain standard knowledge representation, and the quality of the standard knowledge representation is evaluated, and the standard knowledge representation that passes the quality evaluation is used as the knowledge graph.

[0046] Specifically, entity alignment refers to associating and matching entities in different data sources to establish relationships between entities and form a unified knowledge graph. In performing entity alignment, as one possible implementation, the following steps can be included: first, pre-process entity information in different data sources, and extract entity feature vectors from the pre-processed entity information. Data preprocessing is to clean, standardize and uniformly name entities in different data sources, and to perform entity disambiguation and coreference resolution as shown in the background art to eliminate ambiguity of entities with the same name in the text and different expressions of the same entity, and to ensure consistency and standardization of entities. Feature extraction obtains entity feature vectors including word vectors and attribute features, which facilitates subsequent similarity calculation and entity matching. Figure 2

[0047] Then, similarity between different entity information is calculated according to the entity feature vectors, and entity matching is performed according to the similarity. Similarity calculation is to calculate the similarity between entities according to the feature vectors of the entities, and different similarity measurement methods can be used, such as cosine similarity, Jaccard similarity, etc. Entity matching is to filter out entity pairs with high similarity by setting a threshold, and to align and associate entities to establish entity mapping relationship.

[0048] Further, quality evaluation refers to evaluating the effectiveness, accuracy and completeness of the constructed knowledge graph to ensure the quality and credibility of the knowledge graph. In this embodiment, the quality of the standard knowledge representation is evaluated, including entity linking accuracy evaluation, relationship integrity evaluation, query response efficiency evaluation and data updating and maintenance evaluation.

[0049] Among them, entity linking accuracy evaluation refers to evaluating the accuracy and completeness of entity alignment, checking whether the entity matching result is correct, missing and error, etc. Relationship integrity evaluation refers to evaluating the integrity of entity relationships in the knowledge graph, checking whether the entity relationships cover all related information and associations, query response efficiency evaluation refers to evaluating the performance of the knowledge graph in terms of query response speed, accuracy and efficiency, checking whether the knowledge graph can quickly and accurately respond to user queries, data updating and maintenance evaluation refers to evaluating the strategy and effect of knowledge graph data updating and maintenance, checking whether the knowledge graph is updated in time and maintains the accuracy and real-time of data.

[0050] ​Further, the extracted entity, relationship and attribute information in the application is formed into a data layer and a schema layer after evaluation, and is stored in a knowledge base, thereby generating a knowledge graph, so as to form an online operation closed-loop management process according to the stored information in the knowledge base. The data layer can be regarded as the relationship of entity-relation-entity, and the schema layer can be regarded as an ontology library. Forming the schema layer means designing and establishing the logical structure among the entities, relationships and attributes in the ontology library, determining the definition, attributes and relationships of each entity in the ontology library, and the hierarchical structure and constraint relationship among them; and forming the data layer means filling the specific data into the ontology library, that is, storing the data in the ontology library according to the designed schema layer structure.

[0051] The schema layer of the ontology library defines and standardizes each entity, relationship and attribute in the ontology library, and can standardize the relationship between entities in the ontology library to ensure the consistency and reliability of the data. The schema layer also includes metadata information of the ontology library, so that users can better understand and utilize the data in the ontology library. Therefore, the schema layer is the foundation and architecture of the ontology library, and plays an important role in guiding and standardizing the data storage, retrieval and application of the ontology library.

[0052] It should be noted that, as shown in Figure 2 When generating the graph, the standard knowledge representation obtained by entity alignment can also be subjected to knowledge reasoning to discover more related knowledge, so as to increase the amount of obtained knowledge. When performing quality evaluation, the directly collected knowledge and the reasoned knowledge are summarized, and all the knowledge is evaluated to improve the comprehensiveness of the generated knowledge graph.

[0053] In step S102, an online operation platform of after-sales experts is constructed based on the knowledge graph, and the online operation platform is used to answer the after-sales problems raised by users online, and the knowledge graph is updated according to the information generated in the problem answering process.

[0054] Specifically, the online operation platform of after-sales experts is established based on the knowledge of engineers, customers, equipment, software and operation work in the knowledge graph, and the platform has corresponding service ends for different elements. For example, as shown in Figure 3 The platform includes after-sales service engineers, mine users, mining equipment, equipment software and operation work, and the after-sales service engineers include hospital engineers and hospital after-sales service managers and other engineers providing after-sales service.

[0055] In an embodiment of the present application, the online operation and maintenance system answers the after-sales problems raised by the user online, including: receiving the after-sales problems raised by the mine-side user, determining the candidate engineer set corresponding to the after-sales problem based on the knowledge graph, and sending the after-sales problem to each candidate engineer; determining the target after-sales service engineer answering the after-sales problem from the candidate engineer set through the way of claiming or superior supervisor assignment; sending the problem solution feedback by the target after-sales service engineer to the mine-side user for offline processing.

[0056] Further, after sending the problem solution feedback by the target after-sales service engineer to the mine-side user for offline processing, it further includes: obtaining the problem processing result returned by the mine-side user; in the case that the after-sales problem is not solved, sending the problem processing result of this round to the target after-sales service engineer for re-answering, and counting the number of problem answering, in the case that the number of problem answering exceeds the preset threshold, replacing the target after-sales service engineer; in the case that the after-sales problem is solved, obtaining the score of the target after-sales service engineer by the mine-side user, and adjusting the rating of the target after-sales service engineer according to the score and the tracking record of the superior supervisor.

[0057] Specifically, in the present embodiment, the expert online operation and maintenance includes the expert account actually logged in the background, the expert account needs to be matched according to the preset identity verification, if the matching is successful, the expert account can be logged in, wherein the identity verification includes the verification password composed of numbers, symbols and letters, the account can be logged in through the verification password, only the authorized expert account has the right to reply the customer information.

[0058] Specifically, when providing the expert online service to the mine-side user, as shown in Figure 4 The mine-side user can click start in the personal mobile terminal application and raise the problem, the terminal sends to the platform, the operation and maintenance system calls the knowledge graph, obtains the knowledge such as the location of the mine equipment to be processed, the equipment fault type and the fault history answering data from the graph, selects the candidate engineer set capable of processing the problem from all the engineers registered in the platform, and sends the current after-sales problem to each candidate engineer.

[0059] The corresponding hospital-side engineer and hospital-side after-sales service supervisor check the problem, the hospital-side engineer claims the problem according to the self condition (including the self after-sales service ability, the idle time, etc.) after checking the problem, and answers the claimed problem, and the hospital-side after-sales service supervisor tracks the problem answering timeliness after checking the problem, and assigns the problem to a target hospital-side engineer if the problem is not claimed within the preset time length. The mine-side user can check the problem answering process in time after the hospital-side engineer and the hospital-side after-sales service supervisor answer the problem, and performs offline processing and returns the problem processing result. If the problem is not solved, the problem needs to be submitted again, and the hospital-side engineer and the hospital-side after-sales service supervisor answer again. When the hospital-side engineer and the hospital-side after-sales service supervisor answer the same problem for many times, that is, when the number of problem answering times exceeds the preset threshold, it is indicated that the current target after-sales service engineer may not be able to answer the problem, and different answering personnel need to be replaced to improve the fault tolerance rate. The mine-side user performs offline processing according to the answer result, and if the problem can be processed according to the answer, the answer evaluation can be performed and the question and answer are ended, the hospital-side after-sales service supervisor tracks the problem evaluation and records, and the information tracked by the hospital-side after-sales service supervisor includes the number of answers, the user's processing feedback result and the supervisor's evaluation on the answer, etc.

[0060] In the embodiment, a grade score is also set in the service engineer account. After each after-sales question and answer is ended, the system pops up a window to inquire whether the customer's after-sales problem is solved. If it is solved, the expert is scored according to the solution grade, which is divided into the first grade, the second grade and the third grade. If it is not solved, it is also divided into the first grade, the second grade and the third grade. The average grade and the maximum score grade of the scored expert account are displayed on the head portrait side. The customer who needs to answer after-sales can select according to the head portrait grade of the expert account. If the customer thinks that the grade does not meet the after-sales solution demand, the customer can select to replace the expert. The system then offline the expert account and online a new expert account to connect with the customer. When scoring the expert, the evaluation of the mine-side user and the evaluation given by the superior supervisor according to the relevant regulations can be combined to determine the score.

[0061] Therefore, the online operation and maintenance closed-loop management process of the application is from proposing a problem, claiming a problem, assigning a problem to answering a problem, evaluating a service, and converting the knowledge accumulated in the question and answer process into knowledge base data. For example, the most effective processing measures and operation parameters obtained through multiple rounds of answers and actual operation verification of the user are stored in the knowledge base as the corresponding operation and maintenance measures knowledge, which accumulates data for subsequent intelligent question and answer.

[0062] Step S103, constructing a user interest model according to the real-time updated knowledge graph, calculating a user demand similarity matrix according to the user interest model, predicting a user after-sales service demand score according to the user demand similarity matrix, and generating an operation and maintenance knowledge recommendation list for the user according to the after-sales service demand score.

[0063] Specifically, the application can also establish a user interest model through a knowledge graph, score and predict the similarity of user demand, and finally obtain a recommendation list. Based on the recommendation list, the online service system can intelligently ask and answer questions with the user without manual online service of an after-sales expert, and recommend relevant problem-solving operation and maintenance measures and device operation and maintenance knowledge to the user through multiple rounds of question and answer dialogues.

[0064] In an embodiment of the application, the online operation and maintenance intelligent recommendation process of the application is as shown in Figure 5 The target user interest model is constructed according to the related knowledge about the current user to be served in the existing knowledge graph (i.e., the existing knowledge shown in the figure), and the recommendation list for the target user is finally established through the step-by-step operation of the user interest model.

[0065] In the online operation and maintenance closed-loop management process of the application, the actual after-sales knowledge generated by the after-sales expert online operation and maintenance platform can continuously update the knowledge graph. Therefore, the user interest model is constructed according to the real-time updated knowledge graph, including: when the user interest model is initially constructed, if the knowledge graph generated in step S101 has been updated based on the information generated in step S102, for example, the target user has previously performed online consultation of an after-sales expert, the user interest model is constructed according to the updated knowledge graph. In subsequent applications, the parameters in the established user interest model can also be continuously updated and adjusted according to the real-time generated information in step S102, so as to dynamically adjust the user interest model according to the change of the user's interest, and improve the real-time performance and accuracy of the online intelligent after-sales service.

[0066] In order to more clearly illustrate the specific implementation process of constructing the user interest model of the application, a model construction method proposed in an embodiment of the application is exemplarily described below. Figure 6 The flow chart of the user interest model construction method proposed in the embodiment of the application is as shown in Figure 6 The process includes the following steps:

[0067] Step S601, determining the interest field of the user, collecting data related to the interest field from the knowledge graph, and the data is represented as entity information, relationship information and attribute information.

[0068] Specifically, first, a field of interest to be modeled is determined, which can be a work field of the user, including various mining equipment and various underground work to be handled that the user is responsible for. Then, data related to the field is collected, which can be behavior data, preference data, etc. of the user, such as operation data of the user on different mining equipment, historical after-sales service data such as previously consulted operation and maintenance data, operation and maintenance data frequently consulted by the user, and operation and maintenance scheme types preferred by the user, etc. collected from the knowledge graph.

[0069] It can be understood that, since the embodiment is to collect related data from the constructed knowledge graph, and as described above, the knowledge graph in the application is constituted by standardized entity information, relationship information and attribute information, therefore, the data related to the field of interest collected by the embodiment is in the form of entity information, relationship information and attribute information.

[0070] In step S602, the entity information, relationship information and attribute information in the data are represented by vectors by using the TransR model, and a user interest vector is calculated according to the vector representation of the obtained triplets.

[0071] The TransR model is a representation learning model based on relationship embedding, which represents entities and relationships by mapping them into different spaces. The TransR model can map entities and relationships into different vector spaces, and learn the interaction relationship between entities and relationships by defining a transition matrix. The application uses the TransR algorithm to obtain vector representations of entities, relationships and attributes.

[0072] In specific implementation, as a possible implementation manner, the vector representation of the triplets can be performed by the following steps: first, initialize the entity and relationship vectors. The entity and relationship vectors can be initialized randomly or using pre-trained vectors. Then, define the transition matrix. Define a transition matrix for each relationship, which maps the entity vector into the vector space of the relationship. This transition matrix is specific to the relationship, so each relationship has an independent transition matrix. Then, calculate the transitioned entity representation. For a given triplet <head entity, relationship, tail entity>, first, the head entity vector is transitioned by the transition matrix corresponding to the relationship to obtain the head entity representation in the relationship space. Then, calculate the tail entity representation. Calculate the theoretical tail entity vector in the relationship space, which is the difference between the head entity vector and the relationship vector. Finally, minimize the difference between the entity representations. Optimize the entity vector by minimizing the difference between the head entity vector and the transitioned head entity representation.

[0073] Thus, a vector representation of the triplets is obtained, based on which a user interest vector can be calculated, for example, by calculating the weight of each attribute in the set of triplets and performing a weighted sum, it should be noted that since a user can have multiple mining equipment online after-sales service interests, multiple user interest vectors can be obtained. The obtained user interest vector can be used to construct a user interest model.

[0074] Step S603, an initial interest model is created, and the initial interest model is trained by the user interest vector.

[0075] Specifically, according to the parameters of the obtained user interest vector, a suitable modeling algorithm is selected, such as collaborative filtering, content filtering and deep learning algorithm, etc., to construct an initial model. Then the model is trained, and the processed data, i.e. the obtained user interest vector, can be used for model training, for example, a training data set is constructed according to the calculated user interest vector and the related data collected in step S601, and the user interest is used as the output target of the model, the initial interest model is trained, and the trained interest model is obtained.

[0076] Step S604, the trained interest model is evaluated and optimized to generate a user interest model.

[0077] Specifically, the trained interest model is evaluated, and the performance of the model can be evaluated by cross-validation method, and then the model is optimized according to the evaluation result, for example, the parameters and features of the model are optimized.

[0078] Thus, the user interest model is obtained, and the optimized interest model can be applied to the actual scene of the mining equipment online after-sales service, such as a personalized recommendation system, to improve the user experience and satisfaction.

[0079] Further, the user interest model is established based on the knowledge graph to score and predict the similarity of user demand. In an embodiment of the present application, a user demand similarity matrix is calculated according to the user interest model, including: representing the user interest model as a user vector, the user vector is used to determine the similarity between the user and the after-sales service demand; the user demand similarity matrix is calculated by calculating the similarity between the user vectors corresponding to different users.

[0080] Specifically, in the interest model constructed in the present application, the interest of a user can be represented as a vector, and this vector can be used to determine the similarity between the user and the after-sales service demand, and the user demand similarity matrix can be obtained by calculating the similarity between the user vectors. As an example, the cosine similarity between the user vectors can be calculated, and the closer the cosine similarity value is to 1, the more similar the interests of the two users are.

[0081] Further, in the embodiment, the user's after-sales service demand score is predicted according to the user demand similarity matrix, including: predicting the score of the target user for the un-scored after-sales service demand through a weighted average calculation method, wherein the weight in the weighted average is the product of the score of the un-scored after-sales service demand of other different users and the corresponding user demand similarity matrix, and the corresponding user demand similarity matrix is the user demand similarity matrix between other different users and the target user.

[0082] Specifically, there are many methods for predicting the after-sales service demand score according to the similarity matrix, and a classical method is the collaborative filtering method, in which the similarity matrix between users is first calculated, and then the score of the user for the un-scored item is predicted according to the matrix, for example, the score of the user for the item can be predicted by the weighted average method, wherein the weight is the product of the score of the item by other users and the similarity between the user and other users. In addition to the above-mentioned collaborative filtering method, the score can also be predicted by other methods according to the similarity matrix, such as the content-based recommendation method, the model-based recommendation method, etc. The difference between different methods is how to use the similarity matrix and the user interest model to predict the score of the user for the item. In actual application, a method suitable for a specific scenario can be selected to predict the score.

[0083] Further, the operation and maintenance knowledge recommendation list for the user is generated according to the predicted after-sales service demand score. For example, the high and low of each after-sales service demand score can be sorted in descending order to obtain an after-sales service demand sequence, and the after-sales problem solving scheme and related operation and maintenance knowledge corresponding to each after-sales service demand are obtained from the generated knowledge graph, thereby generating the operation and maintenance knowledge recommendation list for the user.

[0084] Step S104, in the case that the user makes an intelligent question and answer request, a multi-round dialogue with the user for the current question and answer request is carried out based on the operation and maintenance knowledge recommendation list.

[0085] In the embodiment, the user can make an intelligent question and answer request about a certain after-sales service to the intelligent question and answer multi-round dialogue library of the mine equipment after-sales online service system through the after-sales expert online operation and maintenance platform in the above embodiment or a specific constructed personalized knowledge service recommendation system. When receiving the intelligent question and answer request of the user, the operation and maintenance knowledge recommendation list of the current user is called, and a multi-round dialogue with the user for the current question and answer request is carried out by using the operation and maintenance knowledge recommendation list.

[0086] For example, when the user raises an intelligent question about the fault of a certain mining equipment, the interest information of the user is obtained from the operation and maintenance knowledge recommendation list, the mining equipment fault operation and maintenance knowledge related to the work field of the user and the equipment operation and maintenance scheme suitable for the user are preferentially selected, and multi-round dialogue interaction is performed with the user, so that the problem of the user can be confirmed as soon as possible and the most suitable equipment operation and maintenance scheme for the user is provided to the user with fewer dialogue rounds. For another example, the personalized knowledge service recommendation system can also be used to regularly push the operation and maintenance knowledge of the mining equipment related to the work of the user to the mobile terminal of the user.

[0087] Based on the knowledge graph, the user interest model data model is constructed, the related data is obtained, and finally the recommendation list has high accuracy, so that the online operation and maintenance intelligent recommendation result is more accurate.

[0088] Therefore, the online operation and maintenance service system is constructed to collect the mine end problems, and functions such as online expert answer, online operation and maintenance intelligent search and knowledge base intelligent recommendation are realized. The closed-loop management of after-sales service is performed, the progress of after-sales service is fed back to the customer in time, the operation and maintenance knowledge base is continuously enriched, the operation and maintenance problems are summarized, and data support is provided for further improving the service quality.

[0089] In summary, the mining equipment online after-sales service method based on the knowledge graph of the embodiment of the present application provides intelligent online operation and maintenance functions based on knowledge graph and deep learning and other artificial intelligence technologies. On the one hand, the model is trained by using massive data, the powerful aggregation information and reasoning ability of the neural network are used, the recommendation system based on the graph neural network and the knowledge graph is designed, and the hit rate of the recommendation can be effectively improved. The knowledge base question and answer based on knowledge reasoning can provide professional technical support to the mine personnel, realize software intelligent question and answer, provide professional knowledge for solving problems independently, which is beneficial to better safety production, timely and rapid solution of daily problems, and provision of standardized after-sales service process. On the other hand, the online operation and maintenance system of the after-sales expert is established, the communication bridge between the customer and the expert is established, the online operation and maintenance service platform of the expert is established, the quality of the after-sales service is improved, and the operation and maintenance data actually generated are summarized and integrated. According to the actual experience data, the model can be further optimized and the operation and maintenance knowledge base can be enriched. Therefore, the method provides an online operation and maintenance closed-loop management process specification, reduces the repetitive operation and maintenance work, provides intelligent online operation and maintenance functions based on knowledge graph and other artificial intelligence technologies, improves the accuracy and timeliness of the after-sales service of the mining equipment, and reduces the service cost.

[0090] In order to realize the above-mentioned embodiments, the present application further provides a mining equipment online after-sales service system based on a knowledge graph, Figure 7 A structure diagram of a mining equipment online after-sales service system based on a knowledge graph according to an embodiment of the present application is shown in Figure 7As shown, the system comprises an acquisition module 100, a first operation and maintenance service module 200, a generation module 300 and a second operation and maintenance service module 400.

[0091] The acquisition module 100 is configured to acquire a plurality of types of knowledge related to after-sales service of the mining equipment, wherein the plurality of types of knowledge comprises a plurality of elements of online operation and maintenance service of the mining equipment, and the plurality of types of knowledge are processed to generate a knowledge graph.

[0092] The first operation and maintenance service module 200 is configured to construct an after-sales expert online operation and maintenance platform based on the knowledge graph, to answer the after-sales problems raised by the user online through the online operation and maintenance platform, and to update the knowledge graph according to the information generated in the process of answering the problems.

[0093] The generation module 300 is configured to construct a user interest model according to the real-time updated knowledge graph, to calculate a user demand similarity matrix according to the user interest model, to predict a score of after-sales service demand of the user according to the user demand similarity matrix, and to generate an operation and maintenance knowledge recommendation list for the user according to the score of after-sales service demand.

[0094] The second operation and maintenance service module 400 is configured to, in the case that the user raises an intelligent question and answer request, carry out a plurality of rounds of dialogues with the user based on the operation and maintenance knowledge recommendation list and the current question and answer request.

[0095] It should be noted that the above description of the embodiment of the method for online after-sales service of the mining equipment based on the knowledge graph is also applicable to the system of the embodiment, and the specific functions performed by each module can be referred to the related description in the above method embodiment, which will not be described here.

[0096] In summary, the online after-sales service system of the mining equipment based on the knowledge graph provided by the embodiment of the present application provides an online operation and maintenance closed-loop management process specification, reduces repetitive operation and maintenance work, provides intelligent online operation and maintenance functions based on artificial intelligence technologies such as knowledge graph, improves the accuracy and timeliness of after-sales service of the mining equipment, and reduces service costs.

[0097] In order to realize the above-mentioned embodiments, the present application further provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method for online after-sales service of the mining equipment based on the knowledge graph as described in any one of the above embodiments.

[0098] In the description of the application, reference to "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that a particular feature, structure, material, or characteristic being described is included in at least one embodiment or example of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment or example. Furthermore, the described specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples. Moreover, the usage of the terms "first", "second" or "third" does not limit the quantity or order of the specific features, structures, materials or characteristics, but rather the term "first", "second" or "third" can be used to distinguish different features, structures, materials or characteristics, which can be combined in any suitable manner. Furthermore, the singular forms "a", "an" and "the" include plural references unless the context clearly dictates otherwise.

[0099] Furthermore, the terms "first", "second", or the like, merely denote different instances of a similar feature, structure, material or characteristic, without necessarily implying any relative importance or any particular order. Thus, a feature defined with "first" or "second" can implicitly or explicitly include at least one of the features. The meaning of "a", "an" and "the" includes plural references unless the context clearly dictates otherwise.

[0100] Any process or method descriptions or blocks in flow charts or otherwise described herein represent embodiments which can be managed as one or more modules, segments, or portions of code which include one or more steps for implementing specific logic functions or steps, and the terms in the description are used for causing or carrying out or upgrading of an action between other hardware under their control. The description of processes and methods of operations should be considered as merely illustrative of the principles of the application, and not in limitation of its scope.

[0101] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a sequence of instructions to implement logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- based system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a computer- readable storage medium or a computer-readable signal medium. The computer- readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium include the following: an electrical connection having one or more wires (electrical connections), a portable computer diskette (a magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, for example, via optical scanning of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.

[0102] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. As such, in some embodiments, specifically configured hardware can be used to implement at least some of the functionality described herein. For example, if implemented in hardware, the hardware can include any or a combination of the following: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0103] Those of skill in the art would understand that information and signals can be represented using any of a variety of technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that can be referenced throughout the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0104] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0105] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A knowledge graph-based online after-sales service method for mining equipment, characterized in that, Includes the following steps: Acquire multiple types of knowledge related to after-sales service of mining equipment, including multiple elements of online operation and maintenance service of mining equipment, and process the multiple types of knowledge to generate a knowledge graph; An online after-sales expert operation and maintenance platform is built based on the knowledge graph. The platform answers after-sales questions raised by users online and updates the knowledge graph based on the information generated during the question-answering process. A user interest model is constructed based on a real-time updated knowledge graph. A user demand similarity matrix is ​​calculated based on the user interest model. The user's after-sales service demand score is predicted based on the user demand similarity matrix. A maintenance knowledge recommendation list is generated for the user based on the after-sales service demand score. When the user makes a request for intelligent question answering, the system will conduct multiple rounds of dialogue with the user regarding the current question answering request based on the operation and maintenance knowledge recommendation list. The step of calculating the user demand similarity matrix based on the user interest model includes: The user interest model is represented as a user vector, which is used to determine the similarity between a user and their after-sales service needs. The user demand similarity matrix is ​​calculated by calculating the similarity between the user vectors corresponding to different users. The step of generating a maintenance knowledge recommendation list for the user based on the after-sales service demand score includes: The after-sales service requirements are sorted in descending order of score to obtain the after-sales service requirement sequence. The knowledge graph is used to obtain solutions to after-sales problems and related operation and maintenance knowledge corresponding to each after-sales service requirement in the after-sales service requirement sequence, thereby generating the operation and maintenance knowledge recommendation list.

2. The method according to claim 1, characterized in that, The process of processing the multiple types of knowledge to generate a knowledge graph includes: Knowledge extraction is performed on the aforementioned multiple types of knowledge to obtain entity information, relation information, and attribute information, and a preliminary knowledge representation is generated based on the extracted information. The preliminary knowledge representation is aligned with entities to obtain a standard knowledge representation, and the standard knowledge representation is evaluated for quality. The standard knowledge representation that passes the quality evaluation is used as the knowledge graph.

3. The method according to claim 1, characterized in that, The online operation and maintenance service includes several elements: after-sales service engineers, mining users, mining equipment, equipment software, and operation and maintenance work. The online operation and maintenance system provides online answers to user-raised after-sales questions, including: The system receives after-sales issues raised by mining users, determines a set of candidate engineers corresponding to the after-sales issues based on the knowledge graph, and sends the after-sales issues to each candidate engineer. The target after-sales service engineer to answer the after-sales problem is determined from the set of candidate engineers by means of claiming or being assigned by the supervisor. The solutions to the problems reported by the target after-sales service engineer will be sent to the mining users for offline processing.

4. The method according to claim 3, characterized in that, After sending the solutions to the problems reported by the target after-sales service engineer to the mining end user for offline processing, the process further includes: Obtain the problem handling result returned by the mining user; If the after-sales problem is not resolved, the result of this round of problem handling will be sent to the target after-sales service engineer for a second answer, and the number of times the problem is answered will be counted. If the number of times the problem is answered exceeds a preset threshold, the target after-sales service engineer will be replaced. Once the after-sales issue has been resolved, obtain the mining user's rating of the target after-sales service engineer, and adjust the target after-sales service engineer's rating based on the rating and the tracking records of the superior supervisor.

5. The method according to claim 2, characterized in that, The step of constructing a user interest model based on a real-time updated knowledge graph includes: Determine the user's interest domain, and collect data related to the interest domain from the knowledge graph, wherein the data is represented as entity information, relationship information, and attribute information; The entity information, relation information and attribute information in the data are vectorized using the TransR model, and the user interest vector is calculated based on the vector representation of the obtained triples. Create an initial interest model and train the initial interest model using the user interest vector; The trained interest model is evaluated and optimized to generate a user interest model.

6. The method according to claim 5, characterized in that, The step of predicting a user's after-sales service requirement score based on the user demand similarity matrix includes: The target user's rating for unrated after-sales service needs is predicted by a weighted average calculation method. The weight in the weighted average is the product of the ratings of other different users for the unrated after-sales service needs and the corresponding user need similarity matrix. The corresponding user need similarity matrix is ​​the user need similarity matrix between other different users and the target user.

7. The method according to claim 2, characterized in that, The entity alignment of the preliminary knowledge representation includes: Entity information from different data sources is preprocessed, and entity feature vectors are extracted from the preprocessed entity information. The similarity between different entity information is calculated based on the entity feature vector, and entity matching is performed based on the similarity. The quality assessment of the standard knowledge representation includes: assessing the accuracy of entity links, the integrity of relationships, the efficiency of query response, and the data update and maintenance of the standard knowledge representation.

8. A knowledge graph-based online after-sales service system for mining equipment, characterized in that, Includes the following modules: The acquisition module is used to acquire multiple types of knowledge related to after-sales service of mining equipment, including multiple elements of online operation and maintenance service of mining equipment, and to process the multiple types of knowledge to generate a knowledge graph. The first operation and maintenance service module is used to build an online operation and maintenance system for after-sales experts based on the knowledge graph, answer after-sales questions raised by users online through the online operation and maintenance system, and update the knowledge graph based on the information generated during the question answering process; The generation module is used to construct a user interest model based on a real-time updated knowledge graph, calculate a user demand similarity matrix based on the user interest model, predict the user's after-sales service demand score based on the user demand similarity matrix, and generate a maintenance knowledge recommendation list for the user based on the after-sales service demand score. The second operation and maintenance service module is used to conduct multiple rounds of dialogue with the user regarding the current question and answer request based on the operation and maintenance knowledge recommendation list when the user makes an intelligent question and answer request. The generation module is also used for: The user interest model is represented as a user vector, which is used to determine the similarity between a user and their after-sales service needs. The user demand similarity matrix is ​​calculated by calculating the similarity between the user vectors corresponding to different users. The generation module is also used for: The after-sales service requirements are sorted in descending order of score to obtain the after-sales service requirement sequence. The knowledge graph is used to obtain solutions to after-sales problems and related operation and maintenance knowledge corresponding to each after-sales service requirement in the after-sales service requirement sequence, thereby generating the operation and maintenance knowledge recommendation list.

9. The system according to claim 8, characterized in that, The acquisition module is specifically used for: Knowledge extraction is performed on the aforementioned multiple types of knowledge to obtain entity information, relation information, and attribute information, and a preliminary knowledge representation is generated based on the extracted information. The preliminary knowledge representation is aligned with entities to obtain a standard knowledge representation, and the standard knowledge representation is evaluated for quality. The standard knowledge representation that passes the quality evaluation is used as the knowledge graph.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the online after-sales service method for mining equipment based on knowledge graphs as described in any one of claims 1-7.

Images